wf_bacterial_prot_src.cwl

#!/usr/bin/env cwl-runner
cwlVersion: v1.2
class: Workflow
label: "Get Proteins"

requirements: 
 - class: SubworkflowFeatureRequirement 
inputs:
 uniColl_asn_cache: Directory
 naming_sqlite: File
 taxid: int
 tax_sql_file: File
 blastdb_dir: Directory
 all_order_specific_blastdb: string[]
outputs:
 universal_clusters:
type: File
outputSource: Get_Proteins_app/universal_clusters
 all_prots:
type: File
outputSource: Get_Proteins_app/all_prots
 selected_blastdb:
type: string[]
outputSource: File2Basenames/values
steps:
 Get_Proteins_app:
label: "Get Proteins: application run"
run: protein_alignment/bacterial_prot_src.cwl
in:
 uniColl_asn_cache: uniColl_asn_cache
 naming_sqlite: naming_sqlite
 taxid: taxid
 tax_sql_file: tax_sql_file
 blastdb_dir: blastdb_dir
 blastdb: 
default: 
 - blastdb
 - blast_rules_db
linkMerge: merge_flattened
 all_order_specific_blastdb: all_order_specific_blastdb
out: [ universal_clusters, all_prots, selected_blastdb ]
 File2Basenames:
# worth describing a trick here:
# Get_Proteins_app/selected_blastdb is a manifest File in Gpipe sense. It contains full absolute paths
# to order specific blastdb 
#
# file paths are transient and unreliable, but information here is in the basename of the selected blastdb, which is 
# a persistent name findable under the unicol reference data
#
# Bottom line: no, we cannot pass manifests from one step to another, we are just using a clever trick here
#
run: progs/file2basenames.cwl
in:
 input: Get_Proteins_app/selected_blastdb
out: [values]